How to Evaluate Shopify Third-Party Integrations: A Scorecard for Apps, Custom Builds and Middleware
A 100-point scorecard for choosing between a Shopify app, a custom build or middleware: data ownership, API health, rate limits, failure handling and three-year cost.
On this page
TL;DR
- Decide data ownership first. Every field, from inventory to customer records, needs one system that owns it. Most integration failures are two systems writing to the same data.
- Check the API before the features. Shopify's REST Admin API has been legacy since October 1, 2024. A connector still built on it carries more risk than one built on GraphQL.
- Rate limits are per app. One middleware connector handling orders, stock and products shares a single limit, which matters at volume.
- Score failure handling, not just success. Ask what happens when a sync fails at 2 a.m.
- Compare three-year cost, not the monthly fee: licenses, per-order charges and maintenance add up differently for apps, custom builds and middleware.
What are your options for connecting Shopify to another system?
There are three: an off-the-shelf app from the Shopify App Store, a custom app built for your store, or middleware (an integration platform) that sits between Shopify and your other systems. Apps are fastest and cheapest to start, custom builds fit unusual processes, and middleware suits businesses connecting several systems that all need to share data.
| Option | Best when | Watch out for |
|---|---|---|
| App from the App Store | A common job, such as syncing to a popular CRM or accounting tool, with standard data | Fixed data mapping, per-order fees, and dependence on the vendor's roadmap |
| Custom app | Your process is genuinely unusual, or data must be transformed in ways no app supports | You own maintenance, including Shopify's quarterly API updates |
| Middleware | Three or more systems (Shopify, ERP, warehouse, CRM) need to share data | Platform license costs, and every flow sharing one Shopify rate limit |
The right answer is usually the simplest one that meets your requirements. As one store owner put it in a verified Clutch review of our work, the value is knowing how "to leverage existing apps to accomplish what I need in my store without having to do custom development."
The integration scorecard
Score each option out of 100 across eight criteria, weighted by how often each causes real-world failures. An option scoring under 60 shouldn't go live without fixes; between 60 and 80 is workable with known risks; above 80 is a strong fit. Score every option the same way so the comparison is fair.
| Criterion | Weight | What to ask | Red flag |
|---|---|---|---|
| 1. Data ownership | 15 |
Which system owns each field, and which way does data flow? |
"Both systems can update it" |
| 2. API health | 15 | Is it built on Shopify's GraphQL Admin API, and on a supported version? | Still relies on the legacy REST Admin API |
| 3. Throughput | 15 | Can it handle your peak order and catalog volume within Shopify's rate limits? Does it use webhooks and bulk operations? | Polls Shopify every few minutes instead of using webhooks |
| 4. Failure handling | 15 | What happens when a sync fails? Are there retries, alerts and logs? | Failures are silent until a customer complains |
| 5. Three-year cost | 15 | What are the license, per-order and maintenance costs over three years? | Per-order pricing with no cap |
| 6. Security and access | 10 |
Does it request only the access scopes it needs? Where is your data stored? |
Requests write access it doesn't need |
| 7. Vendor and exit risk | 10 | Who maintains it, how responsive is support, and can you export your data and mappings? | No way to export configuration if you leave |
| 8. Storefront impact | 5 | Does it add scripts to your storefront, and how do they affect page speed? | Loads heavy scripts on every page |
Why do rate limits matter when choosing an integration?
Shopify limits how much data each app can request, and the limit is set per app and store. On the GraphQL Admin API, standard plans get 100 points per second, Advanced 200 and Plus 1,000. Each update, called a mutation, costs at least 10 points, so large syncs need batching or bulk operations.
A worked example: updating inventory for 5,000 products one change at a time costs about 50,000 points.
| Plan | Points restored per second | Time to finish 50,000 points |
|---|---|---|
| Standard | 100 | About 8 minutes |
| Advanced | 200 | About 4 minutes |
| Plus | 1,000 | Under 1 minute |
Source for rate limits: Shopify's GraphQL Admin API rate limits. Timings are our calculation and ignore batching.
That's fine once a night. It isn't fine if stock needs to update every few minutes during a sale. Well-built integrations use bulk operations and send only what has changed.
The per-app rule cuts both ways. Separate apps don't compete with each other for the limit, but one middleware connector running orders, inventory and products shares a single limit. Ask how a connector schedules competing jobs before peak season, not during it.
Why does the API version matter?
Shopify releases a new Admin API version four times a year and retires old versions on a rolling basis, so every integration needs regular updates. Shopify made its REST Admin API legacy on October 1, 2024, and since April 1, 2025, every new public app has had to use the GraphQL Admin API.
For buyers, that gives two practical questions:
- "Which Shopify API does this use?" GraphQL is the safer long-term answer.
- "How do you keep up with Shopify's quarterly API versions?" A vendor or developer should have a routine, not a reaction.
A custom app that nobody updates is the most common reason a working integration suddenly stops working.
What does good failure handling look like?
A good integration assumes things will fail and makes failures visible. It retries temporary errors automatically, alerts a named person when retries run out, keeps a log you can search, and can re-run a failed sync without creating duplicates. A daily reconciliation report that compares totals between systems catches problems that logs miss.
Five questions worth asking any vendor or developer:
- Who gets alerted when a sync fails, and how?
- Can a failed order or update be replayed safely?
- How long are logs kept, and can we search them?
- Is there a daily check that totals match between systems?
- What happens to orders placed while the integration is down?
How do the three options compare on three-year cost?
Apps usually cost least to start but can cost most at scale if they charge per order. Custom builds cost most upfront but have no license fee, though they need ongoing maintenance for Shopify's API updates. Middleware has a platform license plus setup, and becomes cost-effective when it replaces several separate apps.
When you compare, list for each option over three years: setup cost, monthly license, per-order or usage fees at your projected volume, the cost of keeping up with API changes, and the internal time to manage it. The cheapest month-one option is often not the cheapest over three years.
Integrations we've built
Time4Learning, the online PreK–12 curriculum, needed checkout to capture extra information and pass it to its own learning platform. We built a custom checkout extension that syncs those fields to Time4Learning's internal system as orders come in. Time4Learning case study
HUK Gear runs 59 Shopify POS locations with inventory synced to its ERP, built working directly with HUK's CIO so stock matches across every store and online.
We work with app, custom and middleware approaches, including Celigo, and recommend whichever scores best for your business. See our Shopify app development and ERP, PIM, CRM and WMS integration services.
Shopify integration FAQs
Should I use a Shopify app or a custom integration?
Start with an app if one handles your data and process as they are. Choose a custom build when your process is genuinely unusual, data needs transforming in ways no app supports, or per-order app fees would exceed the cost of building and maintaining your own.
When does middleware make sense for Shopify?
When three or more systems, such as Shopify, an ERP, a warehouse system and a CRM, need to share data. Middleware gives you one place to manage mappings, retries and logs instead of several separate apps.
Is Shopify's REST API being discontinued?
Shopify made the REST Admin API legacy on October 1, 2024, and new public apps have had to use the GraphQL Admin API since April 1, 2025. Existing integrations can still use REST, but new work should use GraphQL.
What are Shopify's API rate limits?
On the GraphQL Admin API, standard plans get 100 points per second, Advanced 200 and Shopify Plus 1,000, applied separately to each app and store. A single query can't exceed 1,000 points.
How often do Shopify integrations need updating?
Shopify releases a new API version four times a year and retires older versions over time, so integrations need a regular update routine. Custom apps that nobody maintains are the most common source of sudden failures.
